Introduction

In-cell NMR has been attracting attention for analyses of the conformations and interactions of proteins in an intracellular environment (Fig. 1). Therefore, in-cell NMR serves to translate in vitro experimental results to in vivo findings. To observe specific protein NMR signals in cells, separately prepared stable isotope-labeled proteins are introduced into cultured mammalian cells by electrical pulse- or toxin-aided cell membrane perforation. Generally, highly concentrated protein solutions (> 1 mM) are required for observations of NMR signals from the introduced intracellular proteins within a reasonable measurement time (< a few hours). Thermal stability at 37°C is also required, as it is the normal growth temperature for cultured human cells. However, only limited numbers of proteins meet these conditions.
　　The aim of this study is to optimize the in-cell NMR method for proteins with poor solubility and low thermal stability. The Chemokine Receptor-Binding Domain (CRBD) of the cytoplasmic chemokine signal-regulator FROUNT [1] (Fig. 2) was utilized.

Methods

All uniformly ¹⁵N-labeled CRBD proteins were prepared with an E. coli expression system. Concentrated protein solutions were mixed with HeLa cells, which were dissociated from cell culture dishes, and introduced into the cells by electrical pulses for cell membrane perforation. For the experiment with the wild-type CRBD protein, highly soluble ¹⁵N-labeled ubiquitin mutant proteins (Ub3A) were also introduced with the CRBD proteins, as a positive control. These cells were cultured on collagen-coated dishes for three hours, to separate attached live cells from unattached dead cells. The live cells were dissociated, washed with phosphate-buffered saline, and mixed with DMEM medium for in-cell NMR experiments. After the experiments, the supernatants of sonicated cells were subjected to NMR measurements to ensure protein structural integrity. NMR experiments were performed with a 14T spectrometer equipped with a TCI CryoProbe (Bruker).

Results and Discussion

Since CRBD protein solutions precipitate at concentrations over 0.8 mM, solutions of 0.6 mM CRBD and an equimolar amount of control Ub3A protein were mixed with HeLa cells, which were subjected to electrical pulses. Collected live cells were used for in-cell NMR experiments. ¹H–¹⁵N SOFAST-HMQC peaks of CRBD, which were observed in vitro (Fig. 3A), were not detected in the in-cell NMR spectrum (Fig. 3B), while those of Ub3A in vitro (Fig. 3C) were observed. Large, self-associated CRBD oligomers were expected to be prevented from entering the cells through the cell membrane pores induced by electrical pulses.
　　To improve the solubility of CRBD, chemical shift differences of ¹H–¹⁵N HSQC peaks were analyzed using CRBD solutions between 0.05 and 0.2 mM to identify the CRBD self-association site. Larger chemical shift differences were observed at a hydrophobic patch surrounded by charged amino acids, based on the protein structure that we recently determined by NMR. This protein surface was predicted to be the self-association site, and thus we designed eleven site-directed mutants on this surface to improve the solubility of CRBD. The prepared mutants were evaluated, based on the NMR peak intensity ratios (0.05 mM vs. 0.5 mM) at 25°C. The peak intensity of the wild-type CRBD was increased by only 4-fold (Fig. 4A), while those of most mutants were increased further. Among them, the peak intensity of the Arg-to-Gln (RQ) mutant was the highest and increased by 10-fold (Fig. 4B), showing that the mutation significantly improved the solubility. Unfortunately, its solubility is high at 25°C, but not at 37°C, the temperature at which HeLa cells normally grow.
　　To overcome this difficult situation, we optimized an in-cell NMR protocol for use at 25°C, from the protein introduction and live cell selection to the in-cell NMR measurement. By using 1.0 mM RQ mutant solutions, ¹H–¹⁵N SOFAST-HMQC peaks, which were observed in vitro (Fig. 3D), were partially observed in the in-cell NMR spectrum at 25°C (Fig. 3E). The observed peaks were from the mobile part of CRBD, and thus the peaks from the structured part were probably broadened due to interactions with large biomolecules. The spectral pattern of the RQ mutant obtained from sonicated cells (Fig. 3F) was identical to that in vitro (Fig. 3D), showing that the three-dimensional structure of the RQ mutant remained properly folded, even though it was subjected to the intracellular environment for three hours.

Conclusion

We successfully optimized the in-cell NMR method for proteins with poor solubility and low thermal stability. By performing experiments according to our strategy, in-cell NMR can be applicable to more types of proteins. Furthermore, we can develop an in-cell system for evaluating therapeutic candidate compounds against target proteins by monitoring the compound–protein interactions.